AI Generated Activities @ Seesaw

Competitive analysis common patterns:

• a large wall of different AI generative tools
• specifc inputs with presets to help get teachers started
• text-only based outputs
• Generic text based quizzes
• Chat Interface for student experience

2) Wireframe & Prototype

I focused ideation on the ingestion experience, uploading a document and entering a prompt, because the structure of the AI-generated output was largely constrained by engineering feasibility and the project timeline. This allowed design effort to be concentrated where it could most meaningfully influence usability, trust, and task success.

I began with low to mid-fidelity wireframes and progressed to mid-fidelity designs to support Figma Make prototypes. This approach enabled rapid sharing, usability testing, and iteration on interaction patterns and flow logic before committing to high-fidelity visuals, ensuring design decisions were validated early and efficiently.

3) Usability Testing

I tested the upload document flow using Figma Make and Lyssna before engineering investment. Because this was a new and unfamiliar experience for teachers, I intentionally spent most of my time validating and refining this part of the flow.

Usability Testing Insights

• Teachers hesitated at a step labeled “Crop,” interpreting it as destructive editing.
• Renaming the action to “Select” and later “Highlight” improved comprehension.
• Added first-time user guidance including a short highlight animation, a “Text Added” badge, and an image counter to reinforce that image selection was the primary action.
• Post-beta engagement data showed low interaction with this step, validating the decision to fully automate image selection from PDFs, which has since shipped.

4) Final UX for Beta Release

For the beta release, we explicitly prioritized learning over optimization. While we were confident this experience would reduce time spent creating activities, the wide variability of curriculum meant that understanding how teachers would prompt and adapt AI-generated activities was a critical learning goal.

• Two distinct creation flows both produced a Seesaw-native, interactive, and auto-graded activity generated by AI.
A lightweight flow allowing teachers to start from a topic or prompt.
A structured flow allowing teachers to upload a document and replicate it as faithfully as possible as a Seesaw activity.
• Prominent, persistent buttons across the top allowed teachers to switch between flows, supporting discoverability and enabling us to observe preference and behavior during beta.